Ciao Leonardo,
I am not sure if this could apply to your case, but maybe - unless you
have done it before - you could look at windowing functions
(http://www.postgresql.org/docs/current/interactive/tutorial-window.html).
They require PG8.4+ though.
Cheers,
Gabriele
On Wed, 4 May 2011 11:51:08 +0100 (BST), Leonardo Francalanci
<m_lists@yahoo.it> wrote:
> Hi,
>
>
> I'm going to need to GROUP BY the same table
> multiple times. That is, something like:
>
> select (some aggregate functions here) from
> tableA group by f1, f2
>
> select (some other aggregate functions here) from
> tableA group by f3, f4
>
> etc
>
> The table is pretty large; can someone suggest the
> best way of doing it? Is running N queries at the same
> time (that is, using N connections with N threads in
> the client code) the only way to speed up things (so
> that the "concurrent scan" thing can help)? Or it's
> more likely that it won't help that much, given that
> we have a fairly good storage? Just trying to get some
> ideas before starting testing....
>
> (table will be 5M rows, where some of the group by
> select could return 3-400K groups)
>
> Leonardo
--
Gabriele Bartolini - 2ndQuadrant Italia
PostgreSQL Training, Services and Support
Gabriele.Bartolini@2ndQuadrant.it - www.2ndQuadrant.it